Figure 2.

Inferences of demographic history and marginal distribution of fitness effects from biallelic data. (A) Biallelic synonymous and nonsynonymous data (thin black lines) and corresponding maximum-likelihood model fits (thick colored lines). Ancestral state misidentification is likely responsible for most of the excess of high-frequency derived alleles, and a parameter to model such misidentification was included in both the synonymous and nonsynonymous models. (B) Inferred demographic model, with two instantaneous population size changes. Time is in units of $2N_{\mathrm{a}}$ generations, where $N_{\mathrm{a}}$ is the ancestral effective population size. (C) Inferred distribution of fitness effects, lognormally distributed for negatively selected mutations with a proportion of positively selected mutations.